Initial support for Korean text to speech
Supports the following:
- allophones (phonemes) for Korean, no long vowels (would require semantic information, barely differentiated in modern Korean), no complex phonemes (such a h morphing) as it is left to the HMM to differentiate them
- hangul (Korean script) to phonemes, with phoneme composition rule (end consonant moves to the next vowel, tensification between patchim consonant and the next, n becomes l after l, etc...)
- hanja (Sino-Korean ideograms) pronunciation, no pronunciation disambiguation (suc as Kim in names becomes keum in nouns, as this would require semantic information)
- support for basic date/time/number expansion
- spells out Latin alphabet text
License:
licensed under LGPL v3.0 with Seunghee loves robots clause, if you use this module on a robot, please send a video with the robot saying a few phrases. Released on her birthday.
The hanja dictionary has been extracted from wikipedia/wiktionary and comes with its own license (same as wikipedia/wiktionary, currently CC-BY-SA, or whatever it becomes).